genetic analysis of benign, low-grade, and high-grade ... · era!blood from the same patient....
TRANSCRIPT
CANCER RESEARCI-155.6172-6180. December 15. 19951
ABSTRACT
Genetic abnormalities were assessed in 56 benign, low-, and high-gradeovarian tumors using comparative genomic hybridization (CGH) andanalysis ofloss of heterozygosity (LOH). In addition, 95 epithelial tumorswere analyzed for microsatellite repeat instability. DNA sequence copynumber abnormalities (CNAs) were not detected in the benign tumors,and more were detected in high-grade than in low-grade cancers. Almost
no microsatellite repeat instability was detected in these cancers. CNAsoccurring in more than 30% of the cancers included increased copynumber on 3q25—26and 8q24 and reduced copy number on 16q andl7pter—q21.Another 14 CNAs occurred in more than 20% of the cancers.Increased copy number at 3q25—26and 20q13 was the most frequent CNAin low-grade tumors, and increased copy number at 8q24 occurred preferentially in high-grade tumors. The presence of a large number of CNAsper tumor was significantly correlated with reduced patient survival
duration. Reduced copy number on l7pter—q21was most strongly maociated with accumulation of a large number of CNAs. The overall concordance between LOH and reduced copy number detected by CGH was84%, but only 31% ofthe LOH was associated with reduced copy numberdetected using CGH.
INTRODUCTION
The ovary is the fifth most common site of the cancer amongAmerican women, and ovarian cancer is the fourth leading cause ofcancer death ( 1). Among patients with pelvic reproductive cancer,more deaths occur from ovarian cancer than from cervical and uterinecancers combined. The survival rate for ovarian cancer varies considerably, depending on the stage at diagnosis. The fractions of patientssurviving 5 years with ovarian cancers diagnosed as Federation Internationale des Gynaecologistes et Obstetristes stage I (limited to theovaries) or stage II (limited to pelvic metastases) are 0.89 and 0.57,respectively. These fractions fall to 0.24 and 0.12 for patients withstage III and stage IV disease, respectively (2). The high mortality rateof ovarian cancer reflects the fact that disease is usually detected late(—70%of all women with common epithelial cancer have stage III orstage IV cancer at the time of diagnosis). Thus, clinical stage is usedroutinely for management of patient treatment. Histological grade isalso used because well-differentiated, low-grade ovarian cancershave better clinical prognoses than do high-grade ovarian cancers(3). However, improved tumor classification is needed becausepatients with tumors that are identical in grade and stage may havesignificantly different clinical outcomes and/or responses totherapy.
One approach to improving diagnosis and/or treatment is to classifytumors according to the genetic abnormalities that they contain. Early
Received 7/19/95; accepted 10/16/95.The costs of publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact.
I This work was supported by Vysis, Inc., the E. 0. Lawrence Berkeley National
Laboratory/University of California, San Francisco Resource for Molecular Cytogeneticsunder United States Department of Energy contract DE-AC-03—76SF00098, and SasakiInstitute Kyoundo Hospital.
2To whom requests for reprints should be addressed, at Division of MolecularCytometry, Department of Laboratory Medicine, MCB 230, Box 0808, University ofCalifomia, San Francisco, CA 94143-0808.
abnormalities may be diagnostically important, whereas informationabout later abnormalities may guide therapy. Genetic changes associated with progression that may be used for tumor characterizationare beginning to be defined for ovarian cancer. Consistent chromosomal abnormalities detected by karyotyping and consistent geneticabnormalities detected in molecular analyses of ovarian cancers arereviewed by Pejovic (4). Regions of frequent LOH3 are summarizedby Osborne and Leech (5) and Yang-Feng et a!. (6). Some events,such as atypical expression or overexpression of egf, fins, and HER2/neu, have been associated with poor prognoses (7—9).However, the
range of genetic abnormalities that are involved in ovarian cancerprogression, their frequency of occurrence, and their clinical andbiological significance remain poorly understood.
In this study, we have further defined the spectrum of geneticabnormalities associated with ovarian cancer, investigated the mechanisms by which the abnormalities occur, assessed their clinicalimportance, and identified abnormalities that may be associated withcancer progression. To accomplish this, we mapped gene dosageabnormalities in tumors of various grades using analysis of LOH (6,10) and CGH (1 i). CGH was particularly useful because it allowedgenome-wide mapping of regions of altered copy number (both increases and decreases from the tumor average) in a single experimentwithout prior knowledge of the locations of regions of abnormality.The comprehensive nature of CGH facilitated correlative analysis ofthe interactions between abnormalities and exploration of the biological and clinical consequences of the various abnormalities.
MATERIALS AND METHODS
Tumor Material. Samples from primary ovarian tumors were obtained
from surgical specimens taken at Yale University Hospital, along with periphera! blood from the same patient. Material was promptly frozen at —70°Cuntilthe time of DNA extraction. Tumor DNA and normal DNA were prepared asdescribed previously (6, 10). All tumors were assigned a histological subtypeand grade. Fifty-six ovarian tumors were analyzed for LOH and copy numberkaryotype using CGH. The cancers analyzed included 26 grade III cancers (21serous cystoadenocarcinomas, 4 endometrioid carcinomas, and 1 mixed epi
thelial carcinoma), 12 grade II cancers (7 serous cystoadenocarcinomas and 5endometrioid carcinomas), and 6 grade I cancers ( I mixed epithelialcarcinoma, 2 serous adenocarcinomas, and 3 mucinous adenocarcinomas).The benign tumors were composed of 10 serous adenomas and 2 mucinousadenomas. Grade I and grade II epithelial ovarian cancers are referred to aslow-grade cancers in this study, and grade III epithelial ovarian cancers arereferred to as high-grade cancers. In addition, 95 epithelial tumors (3benign, 9 borderline, 4 grade I, 14 grade II, and 55 grade III) wereexamined for microsatellite repeat instability. All tumor samples weretrimmed using histological criteria to exclude normal cells as describedpreviously (11, 12).
CGH. CGH was performedusing DNA extractedfrom peripheralbloodlymphocytes and trimmed tumor samples. Isolated tumor DNA was labeledwith biotin-i4-dATP and normal DNA was labeled with digoxigenin-lldUTP. Sixty ng of each of the labeled DNAs plus 5 @gof unlabeled Cot-lDNA were mixed and hybridized to normal metaphase spreads for 3—4days.
3The abbreviations used are: LOH, loss of heterozygosity; CGH, comparativegenomic hybridization; CNA, copy number abnormality.
6172
Genetic Analysis of Benign, Low-Grade, and High-Grade Ovarian Tumors'
Hiroshi Iwabuchi, Masaru Sakamoto, Hotaka Sakunaga, Yen-Ying Ma, Maria L. Carcangiu, Daniel Pinkel,
Teresa L. Yang-Feng, and Joe W. Gra?Department of Laboratory Medicine, Division of Molecular C'ytometry, University of C'alifornia, San Francisco, San Francisco, @alzfornia94143-0808 (H. I., M. S., H. S., D. P.,.1. W. G.J: Department of Gynecology, Sasaki Institute, Kyoundo Hospital, 1-8, Kanda-Surugadai @hiyoda-Ku,Tokyo, 101 Japan [H. 1.. M. S., H. S.]; and Department of Genetics[Y-Y. M., T. L Y-F.J, and Department ofPathology [M. L C.], Yale University, New Haven, Connecticut 06510-8005
Research. on November 1, 2020. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from
GENETIC ANALYSIS OF OVARIAN CANCERS
The preparations were washed to remove unbound DNA and stained withAvidin-FITC (green fluorescence) to detect biotinylated tumor DNA and withanti-digoxigenin-Rhodamine (red fluorescence) to detect digoxigenin-labeled
normal DNA. The chromosomes were counterstained with 4',6'-diamidoinoz-phenylindol for chromosome identification. Three color images from the
metaphase spreads were acquired using a Quantitative Image Processing
System (QUIPS) as described earlier (13). Individual chromosomes were
segmented, local background was subtracted, the medial axes were defined,
and red and green fluorescence intensity profiles were calculated by integratingfluorescence values across the chromosome widths along the medial axes.
Approximately five different metaphases were analyzed for each hybridization.
Green:red fluorescence ratio profiles for four to six chromosomes of the same
type were normalized to standard length and combined statistically to show the
mean and SD of the ratio. A region was considered to be significantlyincreased in DNA sequence copy number relative to the tumor average
when the mean of the green:red ratio was above 1.25 and significantly
decreased in copy number when the mean of the green:red ratio was below
0.75. These excursions are referred to as CNAs. All CNAs were confirmed
1@1/17(6L°'°
by visual inspection of the CGH images. CNAs were not scored for therepeat-rich regions at or near the chromosome centromeres (especiallychromosomes 1, 9, 13—16,21 , and 22) because CGH analyses are unreliablein theseregions.
LOH and Microsateffite Repeat Instability. Tumorswere assessed forLOH at 40 loci distributed as illustrated in Fig. 1. Some loci were studied byanalysis of RFLPs. RFLP analyses were performed as described previously (6).
Briefly, high-molecular weight DNA was isolated from peripheral bloodlymphocytes or ovarian tumors. The DNA was digested using restriction
endonucleases, sized using agarose gel electrophoresis, transferred to a membrane, and hybridized to a 32P-labeledDNA probe. The hybridized probe wasdetected autoradiographically. Quantification of the hybridization signals wasperformed with a laser scanning densitometer. A reduction of more than 50%
relative to the normal signal was recorded as allelic loss.Other loci were assessed for LOH and/or microsatellite repeat instability by
PCR amplification of polymorphic sequences using commercially available
primers (Research Genetics). The PCR reaction mixture consisted of TaqI
DNA polymerase, 200 ,zM dNTP, 1—2mM MgC12, and reaction buffer (Gene
L 0/6(0%)
m 1/19
3/[email protected](19%)
T@13I.(15%)t2/10
(20%)
To'8L(0%)
3/11@ 0/17@JI(0%)(0%)@@ r@27@1 @t4/12@ (0%)
H.L(33%)8
76
(33%) 2/18 @(22%) @1/5 @L(0%)
2/10 1/14 4/18
7/14
(50%) (13%)
. (20%)
20 21 2219
(11%)18
iO/7(0%)
(8%)
x
2/16
1/8
(0%)0/5(0%)5/13(38%)
4/20(20%)
1314[2/16(13%)1/1415
(42%)
Fig. 1. Chromosomal locations of 40 regions tested for LOH. The bounded regions show the areas of the genome to which the polymorphic markers map. The number of loci showingLOH divided by the number of informative loci is shown for each locus.
6173
T1/16I (6%)
T @—5/18(28%)T'@—a'1718%
@2/17(12%)4/14(29%)
‘@—2/14(14%)‘@5/24(21%)
1@4/14@—(29%)
Research. on November 1, 2020. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from
fl:@ Ik@
61@—
@r—@_
@i
fl=@@@@ T :. :•@
@ :::::@::..:::..: .‘@....::::...:: i@@'@
@@ :: ::::::::::::@
@Hm@=@.;@ @;@ :c@j
@ @..:. n@s7@_J__
__@ :::__ @@::::@@
__ ;H@i@
n,@
21.@.
__@@ .::
rt@@-
r:—@@
“:@ ;:::@
‘@l@
GENETIC ANALYSIS OF OVARIAN CANCERS
Fig. 2. Copy number karyotype measured using CGH for a highgrade ovarian cancer. Mean green:red ratios (heavy lines) ± l@(light lines) are shown for each chromosome. , ratios of I .25 and0.75. Bars at the bottom of each chromosome axis show regionsscored as abnormal. Heavy bars, regions of increased relative copynumber; light bars, regions of reduced relative copy number. 7
‘=4
8
Amp kit). Typically, one primer was end labeled with [32P}dATP.PCR wascarried out for 25—35 cycles. The exact thermal cycle profile was adjusted for
each primer pair; usually 30 s to 1 mm at 94°C,30 s to 2 mm at 50—60°C,and30 s to 1 mm at 72°C.An aliquot of the amplified DNA was denatured andseparated on a 6—8% polyacrylamide gel in Tris-borate buffer (0.089 MTris-borate/boric acid-0.002 MEDTA) at 200—300V for 3—6h, depending onthe fragment size. The gel was then exposed to KOdakXAR-5 film for 4—6h.When a microsatellite repeat fragment appeared to be altered in size from that
in the matched normal sample, the experiment was repeated at least twice. In
addition, the autoradiographic films were purposely overexposed to increaseconfidence that the band shift was due to a change in the size of the microsatellite repeat rather than to a PCR artifact.
RESULTS
Commonly Occurring Abnormalities. CNAs were mapped usingCGH in 26 high-grade, 18 low-grade, and 12 benign tumors. A CGHcopy number karyotype and CNA assignments typical of a high-gradetumor are shown in Fig. 2. CNAs were not detected in any of thebenign tumors. Those detected for the low- and high-grade cancers aresummarized and compared in Fig. 3 and Table 1. CNAs appearing in>30%of all cancersarereferredto asclass1abnormalities.
LOH versus CGH. Thirty-nine tumors were analyzed for CNAsusing CGH and for LOH using 40 precisely mapped polymorphicmarkers. In total, 512 loci in the LOH study were informative, and 87showed LOH. These results are summarized in Table 2 and Fig. 1. Not
all regions of the genome were analyzed for LOH with equal resolution. Thus, it is likely that sites of consistent LOH have been missed.The overall concordance between reduced copy number detectedusing CGH and LOH was 0.84. However, the concordance rangedfrom 1.0 to 0.56, depending on the locus. Only 27 of 87 loci showingLOH also showed reduced DNA sequence copy number when analyzed by CGH. Thus, only 31% of the LOH in ovarian cancer isclearly attributable to physical deletions detectable by CGH. Weinvestigated the concordance between CGH and LOH in more detailfor chromosomes i3q and Xp. The results of analysis at 6 loci onchromosome 13q for 34 tumors and analysis of 6 loci on Xp for 35tumors are shown in Fig. 4. The fraction of loci showing both LOHand reduced DNA sequence copy number was 0.65 for chromosomeXp but only 0.08 for chromosome 13q.
Microsatellite Repeat Instability. Ninety-five tumors were analyzed for genetic instability by PCR amplification of 66 differentpolymorphic microsatellite repeat markers. At least one markerwas located on every chromosome arm, except for the short armsof the acrocentric chromosomes. The 66 markers tested werecomposed of 62, 2, and 2 di-, tn-, and tetranucleotide repeats,respectively. In all, 5039 loci were tested in the paired tumornormal samples (not every tumor was tested at each marker). Thenormal and tumor pairs appeared identical in all but six tests,indicating that microsatellite repeat mutations are rare in ovarian
6174
Research. on November 1, 2020. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from
Table I Su@nman of I8 common C'NAs foundin 44 osariancancersHigh
gradeLowgradeCNA
and regionNo.Frequency (%)No.Frequency(%)+
1p22—31831211+
1q25—3172716+
2q32—3372700+
3q25—261350528+5p62316+
6p22—2562300+7q22—3162300—
8p2l—2372700+8q24IS58211+llql4—2283116+l2pl2623317+
13q3l—346232II—16q1038317—
l7pter-q211246317+
18q12—2272716—19727317+20q13519422—Xp7272II
Table 2 Comparison of analyses of LOH and reduced relative DNA copy numberfor512loci in 39 ovariancancersThese
data include 5 cases with LOH and increased relative copy number (scoredasnotdecreased).Decreased
in copy Not decreasedinnumbercopy number TotallociLOH
27 6087NoLOB 20 405425Total
regions 47 465 512
GENETIC ANALYSIS OF OVARIAN CANCERS
.
I
II
.U—
IIU
.U
2
I,@iiI
I@@
IJ@II@ 6
@@ @:i@@@ 11 o@ @:@
kiiit
Fig. 3. Summary of CNAs found during analysis of 44 ovarian cancers. Lines to the left of each chromosome ideogram show regions of reduced relative copy number, and linesto the right show regions of increased relative copy number. . data for high-grade tumors; - , data for low-grade tumors. Each line shows a CNA found in one tumor. CNAswere not scored near the chromosome centromeres because CGH measurements are not reliable in these regions.
tumors, respectively. Table I shows that most of the common geneticabnormalities occurred both in low- and high-grade tumors, but thefrequencies in low-grade tumors were typically less than 20%. Theexceptions were increased copy number abnormalities at 3q26 and20q13, which occurred in 28 and 22%, respectively, of low-gradetumors. Increased copy number abnormality at 8q24 occurred preferentially in high-grade tumors (i.e., an incidence of 58% of high-gradetumors versus 11% of low-grade tumors).
Correlations with Survival Duration. Studies of node-negativebreast cancers have suggested that patients with tumors with manyCNAs have a shorter survival duration than patients whose tumorshave few abnormalities (14). This is consistent with the hypothesisthat tumors accumulate CNAs because they are genetically unstable,and that unstable tumors are able to progress most rapidly. We testedthis possibility for ovarian cancers by assessing survival duration forpatients with >10 and <5 CNAs/tumor and with and without one ormore class 1 abnormalities. The survival durations were shorter for
cancers. Autoradiographs showing the six microsatellite repeatmutations are shown in Fig. 5.
Correlation with Grade. The change in CNA frequency and location with grade was assessed by measuring CGH copy numberkaryotypes for benign, low-, and high-grade tumors. Fig. 6a showsthat the total number of CNAs clearly increases with grade. Specifically, the average total numbers of CNAs per tumor (± 1 u) were0.0 ±0, 5.4 ±7, and 11.2±10for benign,low-, andhigh-grade
6175
@@;i [1 I
@1@I[@ I
Research. on November 1, 2020. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from
GENETIC ANALYSIS OF OVARIAN CANCERS
074 1@@6108 111 112 118 127
@@@ a@@
13 134 1491561581 165
4@169 170 174 175 176 178 186
@@@ •@@
q@@ :187 189 190 197 198 199 202 206
209 2 0 227
•@,@: : DMD
@@ DXS84: • @: DXS7, . TUMP1
DXS255
Fig. 4. Comparison of results obtained using CGH and analysis of LOB on chromosomes l3q (a, 34 tumors) and Xp (b, 35 tumors). Solid lines, regions of reduced relative copynumber detected using CGH; gray lines, regions of increased copy number detected using CGH; 0, regions with no LOH; •,regions with LOH. Loci tested for LOH are shown inthe lower right portions of the panels.
patients with tumors with more CNAs and with one or more class 1abnormality, as illustrated in Fig. 7, a and b. However, the differencewas statistically significant only between patients with and withoutone or more class 1 abnormality (Cox-Mantel; P < 0.05; n = 40).
One important identifier of specific genetic events that contribute todisease progression is correlation between those events and indicatorsof clinical outcome. Correlative studies of the 18 most commonabnormalities detected using CGH in this study identified reducedrelative copy number on 16q as one event associated with reducedsurvival duration (Fig. 7c). However, the association is not strong, andthe number of cases analyzed was small. In addition, because abnormalities at 18 different regions of the genome were tested for correlation with survival, the possibility of finding a correlation by chance
was high. Therefore, this stratification requires validation in a largerstudy.
Correlations between Abnormalities Detected Using CGH.Table 3 shows that there are numerous statistically significant correlations between the common CNAs (especially between the class 1abnormalities). That is, many of the frequently occurring abnormalities seem to occur together in the same tumor. In addition, the class 1abnormalities are associated with a high number of CNAs per tumor.For example, the average number of CNAs per tumor for tumors withone or more class 1 abnormalities is 15.0 ±7.8, whereas that fortumors without any class 1 abnormalities is 0.6 ±0.9 (MannWhitney; P < 0.01). The total number of CNAs is highest for tumorswith reduced copy number involving l7pter—q2l (19.1 ±6.2). Fig. 6,
TN T NTN T N
2
TN
Fig. 5. Autoradiographs showing microsatelliterepeat mutations detected at D5S433 (Lane 1),1.365283 (Lane 2), D7S496 (Lane 3), D10S197(Lane 4), D115901 (Lane 5) AND VWF (Lane 6).T, tumor DNA; N, normal lymphocyte DNA;arrows, mutant alleles.
3 41
6176
(a)(b)069074 106108 111112 118127069gn'
I.,:@ @:130134
149153 156o@ o@158
162 163
o@@ @130
:165
166 167@: 0: @:: : @:
.@@@
186 187 190: : :169
170 175 176 178: 0: : 0:@: : : :
@@ .1@@@ I@
197 198 199 202 206: : : @167
.0@
209:0@
210.:=:D13S33
D13S1RB1D13S2
D13S4D13S3
Research. on November 1, 2020. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from
GENETIC ANALYSIS OF OVARIAN CANCERS
(@)
3&Ill4z 25C.)
:i20I
35. - 3q24@:q2Ã gaIn@@@ 8q24gaIn@ 16qlou 17p-q21loss(b)30@@ c::1@
@ c:D25@@
20T T V
15@@
10 • S
5@ S@
O-9E@-E3E3@@ Highgrade
8
High grade -Lowgrade Benign
UI4zC.)
IU)4zC.)
0
EC
.@0I-
Fig. 6. a, distribution of the total number of CNAsin benign, low-grade, and high-grade ovarian tumors.b, distribution of the class I abnormalities in highgrade ovarian cancers. c, distribution of class 1 abnormalities in low-grade ovarian cancers. 0, does notcarry the specified abnormality; •,does carry thespecified abnormality.
(C)
3q24-q26 gain 8q24 gaIn 16q loss 17P-q21 lou
cD
@D@@
S 8@ 8
@E3X-Low grade
30@
25'
20
15
10
5.
gene dosage. Many, however, may have accumulated by chanceduring clonal evolution of genetically unstable tumors. The CNAsthat do contribute to progression in ovarian cancer are likely to bethose that occur frequently in ovarian and other cancers and/or thatcorrelate with clinical outcome or biological behavior. The class 1abnormalities are of particular interest because of their high preyalence and because they have all been observed at high frequencyin one or more other tumor types. These regions contain severalknown genes that may contribute to development or progression inovarian cancer when differentially expressed. For example, reduced copy number on l7pter—q21 may contribute to disregulation
6177
b and c, shows the association between the total number of CNAs andthe various class 1 abnormalities for the low- and high-grade tumors.
DISCUSSION
Candidate Oncogenes and Tumor Suppressor Genes. Thisstudy of 56 benign, low-grade, and high-grade ovarian tumorsusing CGH and analysis of LOH showed 18 regions of abnormalitythat occurred in more than 20% of ovarian cancers. Some of thesemay contribute to cancer progression through activation of oncogenes, inactivation of tumor suppressor genes, or change in relative
Research. on November 1, 2020. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from
GENETIC ANALYSIS OF OVARIAN CANCERS
phate kinase NM23 (20). Reduced copy number on l6q mayfacilitate inactivation of the cell adhesion molecule E-cadherin(21) and the cell adhesion regulator CMAR (22). Increased copynumber on 8q24 may lead to overexpression of the transcriptionfactor CMYC (23), the SRC family tyrosine kinase LYN (24), theDNA-binding protein MYBLI , and the serine-threonine kinaseMOS (25), whereas increasedcopy number at 3q25—26may affectexpression of EVIl (26) and the tyrosine kinase receptor RYK(27). Of course, the regions defined by CGH typically extend overseveral megabases and contain many genes of unknown function.Thus, definitive identification of the genes affected by these andother less frequent CNAs will require that these regions of abnormality be defined more precisely.
Reduced Copy Number and LOH. We analyzed 39 tumors forCNAs using CGH and for LOH using 40 precisely mapped polymorphic markers to assess the extent to which LOH might be caused byphysical deletions detected using CGH. The overall concordancebetween LOH and reduced copy number determined using CGH was0.84. However, this strong concordance was driven by the largenumber of regions showing no CNA or LOH. The concordance wasonly 0.31 in regions showing LOH. Thus, only this fraction of theLOH in ovarian cancer is clearly attributable to physical deletionsdetected using CGH. This may explain why the spectrum of aberrations defined by analyses of LOH (5, 6) is different from that definedby analysis of CGH. In addition, the concordance varied across thegenome. The concordance between LOH and CGH appears high inregions in which LOH and reduced copy number occur as a result ofextended physical deletions that are reliably detected using CGH. Fig.4a shows that this occurs on Xp. The concordance between LOH andCGH is lower when the regions of LOH are interspersed with regionsthat do not show LOH. Fig. 4b shows that this occurs on chromosome13q.
Genetic Instabilityand CNA Formation.The totalnumberofCNAs may be considered a measure of the degree of genetic instability, using the logic that unstable tumors will accumulate moreabnormalities as they proliferate more than stable tumors. The variation in the number of abnormalities per tumor is remarkable in thisstudy, ranging from 0 to >30 for high-grade tumors and from 0 to >20for low-grade tumors. The detection by CGH of 0 abnormalities inthese tumors does not appear to be an artifact resulting from normalcell contamination, because each tumor of this type showed LOH forat least one locus tested. Genetically unstable tumors also may be
1.0
0.5
0
0
a@I.
101
lull! (a)@ .L...l..L..il I
.L JJ.J CNAs <5
I j IiCNAs>10
1.0
1.0
0.5
0 1 2 3 4 5
Time(years)
J..J.L_.JLUL;
LI! J....I.J.L.IL I 1._I (b)@t No Class 1 ab‘IL II...!
@i I.'
@:@
Class 1 ab
j_J.L..JLU@@ ________________
!@ : LI_I
@ Nolossl6q
t 11
Loss 16q
6
Fig. 7. Kaplan-Meier curves showing survival for patients with ovarian cancer. a,survival for patients with tumors having >10 CNA versus survival for patients with tumorshaving <5 CNAs. b. survival for patients with tumors having one or more class 1abnormalities versus survival for patients with tumors having no class 1 abnormality. c,survival for patients with tumors having reduced copy number at l6q versus survival forpatients with tumors having normal copy number at l6q.
or inactivation of p53 ( I5) and/or a tumor suppressor gene distal top53 (16), BRCAI (17); the GTPase-activating protein NFl (18); theretinoic acid receptor RARA (19); and/or the nucleotide diphos
Table 3 Correlations between the CNAs found in ovarian cancers using CGH
The numbers in the boxes show the significance of the correlations between the 18 most frequent CNAs. The band locations for the CNAs on the various chromosome arms arethe same as those in Table I.
+ Ip + lq + 2q + 3q + 5p + 6p + 7q —8p + Sq + I lq + l2p + l3q — l6q — l7p, q + l8q — 19 + 2Oq —Xp
+lpNAa+lq<0.01NA+2q0.01NA+
3q<0.010.040.01NA+5p0.01NA+
6p0.02+7q0.040.030.04NA—
8p<0.010.010.04NA+
8q0.03<0.01<0.01<0.010.03<0.01NA+
Ilq<0.010.02<0.010.01NA+
12p0.04NA+
l3q0.030.010.010.03NA—
16q0.030.020.02<0.010.02<0.01<0.01NA—
l7p.q<0.01<0.01<0.01<0.01<0.010.020.02<0.010.05NA+
l8q<0.010.010.010.020.03NA—
19<0.01<0.01<0.01<0.01<0.01<0.01<0.010.04NA+20q<0.01NA—
Xp<0.010.020.02<0.010.020.020.010.020.03
NA0.030.04
<0.01
NA
a NA, not applicable.
6178
Research. on November 1, 2020. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from
GENETIC ANALYSIS OF OVARIAN CANCERS
expected to advance more rapidly than more stable tumors. This wasobserved in this study (Fig. 7). Patients whose tumors had manyCNAs survived for significantly shorter periods than patients whosetumors showed few CNAs.
Several genetic abnormalities have already been identified in one ormore tumors that are associated with genetic instability. These includecell cycle checkpoint genes such as p53 (28) and DNA repair genessuch as hMSH2 (29). However, little is known about mechanisms ofinstability or CNA formation in ovarian cancer. We excluded thepossibility that di-, tn-, and tetranucleotide repeat instability contributes to CNA accumulation by analyzing instability at 66 loci in 95tumors. Only six instances of instability were noted. This is somewhatsurprising because microsatellite instability has been observed inovarian cancer cell lines (30). However, our results are consistent withother studies performed on primary tumors (5, 31, 32), suggesting thatthe instability observed in cell lines occurs during culture.
We also searched for specific genetic abnormalities that wereassociated with high numbers of CNAs. Such abnormalities are ofinterest because the regions to which they map may harbor genes thatcontribute to instability and/or CNA formation. The average numberof CNAs was high for all of the class 1 abnormalities (15.0 ±6.8).This raises the possibility that one or more of these abnormalities playa role in the development of genetic instability and/or in accumulationof CNAs in ovarian cancer. The number of CNAs was highest fortumors with reduced copy number at l7pter—q2l (19.2 ±6 for tumorswith reduced copy number in this region versus 3.4 ±5 for tumorswith normal copy number); thus, p53 is implicated as a gene thatcontributes to instability in ovarian cancer. However, it is intriguingthat essentially all abnormalities involving chromosome 17 extendedfrom l7pter to l7q2l (see Fig. 3). Thus, other genes in this region mayplay a role as well [e.g., a second gene distal to p53 (33) and/or nearBRCAJ (17)1.
Measuresof Progression.Weanalyzedabnormalitiesinbothlowand high-grade tumors in an effort to determine the extent to which tumorgrade has a genetic basis and to identify possible early abnormalities (i.e.,those that occur in low-grade tumors). Analysis of abnormalities intumors differing in stage might be preferable for identification of earlyabnormalities. However, low-stage tumors were not available for thisstudy. Therefore, tumor grade was taken as a measure of the extent oftumor evolution in this study. The higher-grade tumors do seem to bemore advanced genetically than the lower-grade and benign tumors basedon total CNA count (0.0 ±0 versus 5.4 ±7 versus 11.2 ±10 for benign,low-, and high-grade tumors, respectively). Aberrations that occurredmost frequently in low-grade tumors were increased copy number on3q25—26(28%) and increased copy number at 20q13 (22%). It was ofinterest that amplification of sequences at 20ql3 also has been found inbreast cancer and is associated with poor clinical outcome (34), and
increased copy number at 3q25—26has been reported in lung cancer (35).Thus, overexpression of genes in these regions (e.g. , due to amplification)may play a role in early tumor formation. If so, analysis of proteins codedfor by genes in these regions may be diagnostically useful. Increasedcopy number on 8q24, on the other hand, occurs preferentially in highgrade tumors (i.e., 58% incidence in high-grade tumors versus 11%incidence in low-grade tumors). Thus, increase in DNA sequence copynumber at 8q24 may be a late event. Of course, it is important to keep inmind that grade is not a strong indicator of degree of progression inovarian cancer (survival durations of low- and high-grade tumors in thisstudy were not significantly different). Thus, the genetic differencesbetween low- and high-grade tumors detected in this study also may beevents that alter cell morphology and/or organization (e.g., ploidy, rate ofproliferation, or apoptosis).
6179
ACKNOWLEDGMENTS
The authors gratefully acknowledge the continuing support of Drs. Y.Tenjin and T. Sugishita.
REFERENCES
I. Phyllis, A., Tony, 1., and Sherry, B. Cancer statistics. CA Cancer i. Clin., 45: 8—30,1995.
2. Nguyan, H. N., Averette, H. E., Hoskins, W., Sevin, B. U., Penelver, M., and Steren.A. National survey of ovarian carcinoma. VI. Critical assessment of current lnternational Federation of Gynecology and Obstetrics staging system. Cancer (Phila.), 72:3007—3011,1993.
3. Heintz, A. P. M. Surgery in advanced ovarian carcinoma: is there proof to show thebenefit? Eur. J. Surg. Oncol., 14: 91—99,1988.
4. Pejovic, T. Genetic changes in ovarian cancer. Ann. Med., 27: 73—78,1995.5. Osborne, R., and Leech, V. Polymerase chain reaction allelotyping of human ovarian
cancer. Br. J. Cancer, 69: 429—438,1994.6. Yang-Feng. T. L., Han, H., Chen, K. C., Li, S. B., Claus, E. B.. Carcangiu, M. L.,
Chambers, S. K., Chambers, J. T., and Schwartz, P. E. Allelic loss in ovarian cancer.Int. 3. Cancer, 54: 546—551, 1993.
7. Berchuck, A., Kamel, A., Whitaker, R., Kems, B., Olt, G., Kinney, R., Soper, i. T.,Dodge, R.. Clarke-Pearson, D. L., and Marks, P. Overexpression of HER-2/neu isassociated with poor survival in advanced epithelial ovarian cancer. Cancer Rca., 50:4087—4091.1990.
8. Borrensen, A. L. Oncogenesis in ovarian cancer. Acts Obstet. Gynecol. Scand., 155:25—30,1992.
9. Scwab, M., and Amler, L. C. Amplification of cellular oncogenes: a predictor ofclinical outcome in human cancer. Genes Chromosomes & Cancer, 1: 181—193,1990.
10. Yang-Feng, T. L., Li, S., Han, H., and Schwartz, P. E. Frequent loss of heterozygosityon chromosomes Xp and l3q in human ovarian cancer. Int. J. Cancer. 52: 575—580,1992.
II. Kallioniemi, A., Kallioniemi. 0. P., Sudar, D., Rutovitz, D., Gray, 3. W., Waldman,F., and Pinkel, D. Comparative genomic hybridization for molecular cytogeneticanalysis of solid tumors. Science (Washington DC), 258: 818—821, 1992.
12. Kallioniemi, 0. P., Kallioniemi, A., Sudar, D., Rutovitz, D., Gray, 3. W., Waldman,F., and Pinkel, D. Comparative genomic hybridization: a rapid new method fordetecting and mapping DNA amplification in tumors. Semin. Cancer Biol., 4: 41—46,1993.
13. Piper, J., Rutovitz, D., Sudar, D., Kallioniemi, A., Kallioniemi, 0-P., Waldman, F.,Gray. J., and Pinkel. D. Computer image analysis of comparative genomic hybridization. Cytometry, 19: 19—26,1995.
14. Isola, J., Kallioniemi, 0-P., Chu, L., Fuqua. S., Hilsenbeck, K., Osbome, K., andWaldman, F. Genetic aberrations detected by comparative genomic hybridizationpredict outcome in node-negative breast cancer. Am. .1.Pathol., 47: 905—911, 1995.
15. Okamoto, A., Sameshima, Y., Yokoyama, S., Terashima, Y.. Sugimura, T., Terada,M., and Yokota, 3. Frequent allelic losses and mutations of the p53 gene in humanovarian cancer. Cancer Rca., Si: 5171—5176,1991.
16. McDonald, .1., Daneshvar, L., Willert, 3., Matsumura, K., Waldman, F., and Cogen,P. H. Physical mapping of chromosome l'1pl3.3 in the region of a putative tumorsuppressor gene important in medulloblastoma. Gemonics, 23: 229—232,1994.
17. Miki, Y., Swensen, J., Shattuck-Eldens, D., Futreal, A., Harshman, K., Tavtigian, S.,Liu, Q., Cochran, C., Bennett, M., Ding, W., Bell, R., Rosenthal, J., Hussey, C., Tran,T., McClure, M., Frye, C., Battier, T., Phelps, R., Haugen-Strano, A., Katcher, H.,Yakumo, K., Gholami, Z., Dayananth, P., Ward, J., Tonin, P., Narod, S., Bristow, P.,Norris, F., Helvering, L., Morrison, P., Rosteck, P., Lai, M., Barrett, C., Lewis, C.,Neuhausen, S., Cannon-Albright, L., Goldgar, D., Wisemann, R., Kamb, A., andSkolnick, M. Isolation of BRC'AI,the 17q-linked breast and ovarian cancer susceptibility gene. Science (Washington DC), 266: 66—71,1994.
18. Jhanwar, S., Chen, Q., Li, F., Brennan, M., and Woodruff, i. Cytogenetic analysis ofsoft tissue sarcomas. Recurrent chromosome abnormalities in malignant peripheralnerve sheath tumors (MPNST). Cancer Genet. Cytogenet., 78: 138—144,1994.
19. Futreal, P., Soderkvist, P., Marks, J., Iglehart, J., Cochran, C., Barrett, J. C., andWisemann, R. Detection of frequent allelic loss on proximal chromosome llq insporadic breast carcinoma using microsatellite length polymorphisms. Cancer Res.,52: 2624—2627,1992.
20. Cawkwell, L., Lewis, F. A., and Quirke, P. Frequency of allele loss of DCC, p5.3, RB!,WI,!, NFl, NM23 and APC/MCC in colorectal cancer assayedby fluorescent multiplex polymerase chain reaction. Br. J. Cancer. 70: 813—818,1994.
21. Mareel. M., Bracke, M., and van Roy, F. Invasion promoter vs invasion suppressormolecules: the paradigm or E-cadherin. Mol. Biol. Rep., 19: 45—67,1994.
22. Koyama, K., Emi, M., and Nakamura, Y. The cell adhesion regulator (CAR) gene,TaqI and insertion/deletion polymorphisms, and regional assignment to the peritelomeric region of 16q by linkage analysis. Genomics, 16: 264—265, 1993.
23. Baker, V. V., Borst, M. P., Dixon, D., Hatch, K. D., Shingleton, H. M., and Miller,D. c-myc amplification in ovarian cancer. Gynecol. Oncol., 38: 340—342,1990.
24. Corey, S., and Shapiro, D. N. Localization of the human gene for SRC-related proteintyrosine kinase LYN to chromosome 8qI 1—12:a lymphoid signaling cluster? Leukemia (Baltimore), 8: 1914—1917,1994.
25. Dolcetti, R., Pelucchi, A., Maestro, R., Rizzo, S., Pastore, A., and Boiocchi, M.Proto-oncogene allelic variations in human squamous call carcinomas of the larynx.Eur. Arch. Oto-Rhino-Laryngol., 248: 279—285,1991.
26. Morishita, K., Parganas, E., Douglass, E. C.. and IhIe, J. N. Unique expression of thehuman Evi.1 gene in an endometrial carcinoma cell line: sequence of cDNAs andstructure of alternatively spliced transcripts. Oncogene, 5: 963—971,1990.
Research. on November 1, 2020. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from
GENETIC ANALYSIS OF OVARIAN CANCERS
32. Han, H-i., Yanagisawa. A., Kato, Y., Park, J-G., and Nakamura, Y. Genetic instabilityin pancreatic cancer and poorly differentiated type of gastric cancer. Cancer Res., 53:5087—5089,1993.
33. Theile, M., Hartmann, S., Scherthan, H., Arnold, W., Deppert, W., Frege, R., Glaab,F., Haensch, W., and Schemeck, S. Suppression of tumorigenicity of breast cancercells by transfer of human chromosome 17 does not require transferred BRCAJ andp53 genes. Oncogene, 10: 439—447, 1995.
34. Tanner, M. M., Tirkkonen, M., Kallioniemi, A., Collins, C., Stokke, T., Karhu, R.,Kowbel, D., Shadravan, F., Hintz, M., Kuo, W-L., Waldman, F. M., Isola, J. J., Gray,J. W., and Kalliomeini, 0-P. Increased copy number at 20ql3 in breast cancer:defining the critical region and exclusion of candidate genes. Cancer Res., 54:4257—4260,1994.
35. Brzoska, P. M., Levin, N. A., Fu, K. K., Kaplan, M. J., Singer, M. I., Gray. J. W., andChristman, M. F. Frequent novel DNA copy number increase in squamous cell headand neck tumors. Cancer Res., 55: 3055—3059,1995.
27. Simoneaux, D., Fletcher, F., Jurecic, R., Shilling, H., Van, N., Patel, P., andBelmon, J. W. The receptor tyrosine konase-related gene (ryk) demonstrateslineage and stage-specific expression in hematopoietic cells. J. Immunol., 154:1157—1166,1995.
28. Hartwell, L., and Kastan, M. Cell cycle control and cancer. Science (WashingtonDC), 266: 1821—1828,1994.
29. Modrich, P. Mismatch repair, genetic stability, and cancer. Science (Washington DC).266: 1959—1960,1994.
30. Orth, K., Hung, 3., Gazdar, A., Bowcock, A., Mathis, J., and Sambrook, J. Geneticinstability in human ovarian cancer cell lines. Proc. NatI. Aced. Sci. USA, 91:9495—9499,1994.
31. Wooster, R., Cleton-Jensen, A-M., Collins, N., Mangion, J.. Cornelis, R., Cooper, C.,Gusterson, B., Ponder, B., von Deimling, A., Wiestler, 0., Cornelisse, C., Devilee, P.,and Stratton, M. Instability of short tandem repeats (microsatellites) in humancancers. Nat. Genet., 6: 152—156,1994.
6180
Research. on November 1, 2020. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from
1995;55:6172-6180. Cancer Res Hiroshi Iwabuchi, Masaru Sakamoto, Hotaka Sakunaga, et al. Ovarian TumorsGenetic Analysis of Benign, Low-Grade, and High-Grade
Updated version
http://cancerres.aacrjournals.org/content/55/24/6172
Access the most recent version of this article at:
E-mail alerts related to this article or journal.Sign up to receive free email-alerts
Subscriptions
Reprints and
To order reprints of this article or to subscribe to the journal, contact the AACR Publications
Permissions
Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)
.http://cancerres.aacrjournals.org/content/55/24/6172To request permission to re-use all or part of this article, use this link
Research. on November 1, 2020. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from